Last updated: August 27, 2026

How windows get opened, filtered and priced as popunder ads

Impressions begin with a human action. A browser will not open a background window on its own initiative, so the publisher's script waits for a click, a tap or a keypress somewhere on the page and uses that gesture as permission, which means every unit you pay for is tied to a real interaction rather than a page load. That constraint explains what follows. Volume depends on how often visitors touch the page, not on how many pages they open, and pricing for popunder ads follows from exactly that.

A browser gesture is what actually launches popunder ads

No gesture, no window. Pages where visitors read passively produce almost nothing regardless of traffic, while pages built around interaction generate several opportunities per visit, and that difference matters more than audience quality when you compare sources of popunder ads.

Blocked spawns are rarely reported. Browsers refuse windows that arrive too long after the gesture, extensions intercept a further share, and some mobile browsers restrict the behaviour outright, so the gap between attempted and delivered can be substantial. Ask for the delivery rate. Sellers without one report attempts. Sellers reporting attempts rather than deliveries are not hiding anything, but the number quoted will never appear in your own account.

What counts as an impression

The window opening is the billable event on most arrangements, not the page finishing its load. Those are different moments separated by whatever the connection is doing, which is why slow markets show a wider gap between what you paid for and what your own analytics eventually recorded.

Frequency capping decides whether popunder ads earn or simply annoy

One per visitor per day. The first window in an afternoon carries whatever novelty this format has. Everything after it spends goodwill belonging to the publisher rather than to you, which is what buyers forget when raising a cap to hit a number. That single setting does more for the numbers than any targeting refinement advertising platforms offer, because the second window a person receives in an afternoon converts far worse than the first, the third converts worse again, and uncapped delivery produces exactly the reputation that makes publishers restrict popunder ads in the first place.

Capping needs an identity to work against, and identity is the weak point here, because cookies vanish in private windows, storage clears, and fingerprints drift across browser updates, so a cap of one is really one per identifiable session rather than per person. Treat the cap as a quality floor rather than a volume ceiling, because the version that fails is the one loosened quietly in the last week.

Where the cap should live

Set it at the zone. A visitor moving between three sites you are buying on will otherwise receive three windows while every individual cap reports itself as respected, which is technically true and practically useless. The visitor sees three windows, your report sees three compliant campaigns, and the publisher sees one irritated person installing a blocker tonight. Nobody wins that exchange. Plan the zone list accordingly. Central identity is the only real fix, and where it does not exist the zone list itself has to do the work instead. Overlap is a planning problem, not a filter.

Cross-zone capping requires the seller to hold identity centrally and many cannot, so the workable substitute is to avoid buying several zones with heavily overlapping audiences at once, which is routinely ignored by anyone building a zone list from a volume report. Overlapping audiences are easiest to spot by comparing the distinct address counts of two zones against their combined total, which takes ten minutes.

Session dedupe quietly removes a third of the volume in popunder ads

Promised volume is gross. Every one of those layers is defensible on its own and the compounding is what surprises people, because four independent reductions of fifteen percent do not remove sixty percent, they remove closer to half, and the forecast was built on none of them. Between the seller's estimate and your delivered total sit blocked spawns, frequency caps, cross-zone deduplication and invalid traffic filtering, each removing a slice, and the cumulative effect regularly turns a confident forecast into something visibly smaller once you begin buying popunder ads seriously.

None of that is dishonest, and all of it is predictable if you ask the right question before committing. Ask what it measures. An estimate based on gross opportunities and a delivery based on capped, filtered, deduplicated windows are different quantities wearing the same unit, and the ratio between them is stable enough per seller to be learned within a fortnight and then relied upon.

Build the discount into the plan

Measure your own delivery ratio during the first week and apply it to every forecast afterwards. Ratios differ again when you buy adult web traffic. It is usually consistent per seller, it rarely improves on its own, and planning against gross figures guarantees a monthly shortfall you will spend the last week of every month trying to buy your way out of at worse prices. Applying a measured ratio to a seller's estimate takes seconds and removes an entire category of monthly disappointment, which is a good return on a week of careful record keeping. Do it once per source. Forecast against it.

Bot filtering matters more here than anywhere else in popunder ads

Low intent is easy to fake. Cheap Website Traffic says the same thing. A background window requires no engagement to look legitimate, no scroll, no dwell and no considered click, which makes this format the cheapest place in the market to manufacture plausible-looking activity and therefore the place where filtering earns its cost most clearly across popunder ads inventory.

The useful checks are ratios. Compare windows delivered against distinct addresses, sessions against devices, and delivered volume against the publisher's own reported page views, because a zone producing more windows than its audience could plausibly generate is answering the question for you. Ratios. Absolute thresholds go stale within a month. Ratios expose porn traffic zones too. A zone where the address count barely moves as volume triples is answering the question without needing any conversion data at all, and the check costs nothing beyond an export and a spreadsheet column. Always.

The three-day audit

Buy a small volume across a wide zone list, then compare the distribution of session durations by zone rather than the conversion figures. Real audiences produce a long, messy spread, and manufactured ones produce a suspiciously narrow one, which shows up long before any conversion data becomes statistically meaningful. Stale.

I started running that audit after reading through the zone-quality material on popunder-ads.io, and it now precedes every new source I test, because three days of small spend has repeatedly saved a month of slow, expensive disappointment that would otherwise have been blamed on the offer. Conversion figures need weeks to become meaningful and session shape needs days, which is why the audit runs on shape. Real audiences are untidy in a particular way, with a long tail of short visits and a scattering of very long ones, and manufactured ones rarely reproduce that mess.

Where popunder ads sit next to other formats in a media plan

This format buys reach, not consideration. Reach. It reaches people who never asked for anything, at prices per thousand far below any placement that requires attention, and that combination makes it a poor closing instrument and an unusually efficient one for the earliest stage of a funnel where popunder ads do their real work.

Judge it against the right benchmark. Comparing a background window to a search click on cost per acquisition is comparing two entirely different jobs, and the honest comparison is against other volume formats on cost per thousand qualified sessions rather than on final conversion. Judged as a closer it looks poor, judged as the cheapest way to put an offer in front of a large number of unaware people it looks completely different, and the benchmark you choose decides which of those two conclusions you reach. Pick the benchmark first.

What pairs well with it

Retargeting. The value of cheap reach is mostly the audience it builds for a second, more expensive touch, so the plan should assume that first contact rarely closes anything by itself. A first touch costing a fraction of a search click need not close anything to earn its place, provided the second touch exists and somebody watches it.

Those four levers are unglamorous, entirely learnable and almost never the thing a struggling buyer chooses to work on. Caps and zone lists rarely feel like progress while you are maintaining them. Native ads reward the same habits. Volume formats reward operational discipline far more than creative brilliance, because the levers that actually move results are caps, zone lists, delivery ratios and filtering rather than artwork, and a buyer who maintains those four things carefully will beat a more imaginative competitor who ignores them every time popunder ads appear in a plan.